|
PureBytes Links
Trading Reference Links
|
At 11:44 AM 9/24/98 -0700, David Powell wrote:
<snip>
>Please, could someone define these terms so that we may know what the
>differences are?
>
>
>NN
>NN by Genetic Algo
>Backpercolation NN
>NN algorithms
>Kohonen Nets,
>recursive nets, GRNN and PRNN Nets
>NN evolved by GA
>multidimentional Nets, Scorecards, Rule bases, all evolved by very
>sophisticated GA
>neurofuzzy logic
>RMSE error
>evaluation function (Netprofit or RMSE,
>maybe others).
>NN evolved by GA.
>
>-an average (nn challenged) guy
A neural network (more properly Artificial neural network) is a computer
construction (could be hardware, most commonly software) inspired by the
neural network structure of the nervous system. Neural networks come in a
large variety of flavors, some of which are more appropriate to certain tasks.
Neural networks can learn to recognize patterns. If the patterns are
representative and the learning is done well, neural networks can be used
to make predictions. They can also learn to classify data, which can be
used to make decisions. The entries on your list have to do with particular
flavors of neural nets or particular strategies for training them.
Artificial intelligence is considered to include neural networks, genetic
algorithms (inspired by genetic theory and survival of the fittest /
mutation / etc. -- they can be used to find optimal solutions efficiently)
and expert systems. For an introductory discussion of neural networks, you
can try
http://www.wardsystems.com/
and follow the link to About Neural Networks. Some other vendors of neural
network software also have informative web sites.
Allan
"The highest result of education is tolerance." - Helen Keller
|